Slides da palestra "Direto dos Laboratórios da Rakuten Japão: O Futuro do Comércio Eletrônico" feita pelo Yoichi Yoshimoto, Cientista do RIT- Instituto de Tecnologia da Rakuten.
3. Growing Data in Rakuten
Ichiba GMS (∝ # of transaction)
160 M
# of items # of reviews
4. …will be happening in Brazil as well
Source: eMarketer Jan 2014 “Retail Ecommerce Sales in Brazil to See Double-Digit Growth This Year”
5. Growing services with Growing
Data
Vol.01 Oct/14/2014
Yoichi Yoshimoto | Mario
Rakuten Institute of Technology, Rakuten Inc.
http://rit.rakuten.co.jp/
6. • Yoichi Yoshimoto
• Lead Coordinator
Rakuten Institute of Technology
Rakuten, Inc.
• So what’s my role?
Connecting R&D projects to business and tech teams
Currently focusing on “Data Mining” and “Natural
Language Processing” areas.
19. 【Keyword Trend】Peak Season Identification
Re-discovering peak season from time series data
Jan. 1st
Dec. 31st
“School Bags” have 2 peak seasons
Grandparents start looking for
school bags as gifts for their
grandchildren after their visit
during summer vacation.
Maybe!
23. 【Keyword Trend】Discovering event related demands
Burst keywords after Great East Japan Earthquake
We can see demands that aren’t reflected in POS data.
28. Attribute Extraction
Item pages in Rakuten are created by merchants
-> They Contain lots of unstructured text.
For better service, we need structured data.
Hard to see wine’s attributes Easy to see wine’s attributes
29. Attribute Extraction
Item pages in Rakuten are created by merchants
-> They Contain lots of unstructured text.
For better service, we need structured data.
Hard to see wine’s attributes Easy to see wine’s attributes
30. Attribute Extraction
We can extract attributes regardless of categories or
languages as long as table data is available
32. GEAP: Global Event Analysis Platform
• Collecting log data from any devices and services
with single platform.
• Big data analysis of cross services